Structured Query Language SQL I

History • IBM – after 1970 - relational DBMS prototype - System R • After 1908 – basis of 2 commercial DBMSs: SQL/DS, DB2

SQL standard

• Standardization institutions

• ANSI: American National Standards Institute • ISO: International Organization for Standardization • IEC: International Electrotechnical Commission

• SQL86 (sometimes called SQL87) - SQL 1 ! 1986 ANSI X3.135-1986 Database language SQL ! 1987 ISO 9075-1987 Database language SQL Referential integrity not standardized before 1989. ! 1989 ANSI X3.135-1989 Database Language SQL With

Integrity Enhancement ! 1989 ISO 9075-1989 Database language SQL

History II

! Embedded SQL

! 1989 ANSI X3.168-1989 Database Language Embedded SQL ! neexistuje ISO standard pro embedde SQL

Embedded SQL allows for asking SQL queries from a program written in classical host programming language, typically C language

Queries written directly to C source code.

Special preprocessor replaces these queries with invoking respective functions that are part of DBMS vendor’s libraries.

Around 1990 frequently used when developing database applications in C languege.

History

! SQL92 – aka SQL2: • 1992 ANSI X3.135-1992 Database language SQL ! 1992 ISO/IEC 9075-1992 Database language SQL

• The most frequently used SQL standard. • These lectures will focus on SQL 92

Further development

! SQL99 – aka SQL3 ! Regular expressions ! Recursive queries ! Non-scalar data types ! Etc.

! 2003: SQL2003 ! XML support ! Standardized concept of SEQUENCE and automatically generated values

http://www.wiscorp.com/sql_2003_standard.zip

! 2006: SQL2006 ! Extended support of XML (XQuery) ! Conversions XML ↔ SQL ! Export/import XML

SQL – data types I

Numerical data types stored precisely (fixed decimal point)

INTEGER integers Range implementationally dependent, typically between -2147483648 and 214783647

SMALLINT integers Range implementationally dependent, typicallye -32768 až 32767. Definition: range not bigger than INTEGER.

NUMERIC NUMERIC(p) NUMERIC(p,s)

Number – possibly with fractional part. Decadic number with p digits, s of them behind decimal point.

E.g. DECIMAL(5,2) has 3 positions on the left of the decimal point and 2 positions on the right of the decimal point. If p or s not expressed explicitly, a default is use that is dependent on particular implementation.

The number is always stored with the respective precission (unless a „physical“ limit reached)

DECIMAL DECIMAL(p) DECIMAL(p,s)

Similar to NUMERIC. However, the number may be stored more precisely that prescribed (unless „physical“ limit reached).

Remark: NUMERIC – may be stored as string of digits, while DECIMAL in interbal (binary) representation of the respective processor type.

SQL – data types II

Numeric data types stored imprecisely (floating decimal point) REAL Floating decimal

point – simple precission

Precision implementationally dependent Usually default precission for data with floating decimal point on given HW platform

DOUBLE PRECISON

Floating decimal point – double precission

Precision implementationally dependent Usually double precission for data with floating decimal point on given HW platform Definition: precision greater than REAL.

FLOAT FLOAT(p)

Allows for defining the requested precission. The number may be stored with higher precision than prescribed.

The requested precision p may be achieved with simple (default) precision on one platform, whereas with double precision on another platform.

SQL – data types III

Strings of characters

CHARACTER CHARACTER(x)

String of characters with specified length. If x not expressed, equiv. To CHAR(1).

CHARACTER VARYING VARCHAR CHARACTER VARYING(x) VARCHAR(x)

Strings wit variable length (alocated so many bytes as needed for particular string). The maximal length limited by x. Maximal length depends on particular implementation.

NATIONAL CHARACTER NCHAR NVARCHAR

Support for national alphabets. UNICODE

SQL – data types IV

Date and time DATE Date:

Length 10 chars incl. separators YYYY-MM-DD

TIME TIME(p)

Time: p number of decimal positions (fraction of a second) Length 8 characters if p=0, else 9+p Default 0 decimal positions HH:MM:SS HH:MM:SS.PPP

TIMESTAMP TIMESTAMP(p)

Date + time. Length 19 positions if p=0, else 20+p Default 6 decimal positions YYYY-MM-DD HH:MM:SS

CREATE TABLE I CREATE TABLE PACKAGE ( PACKID CHAR(4), PACKNAME CHAR(20), PACKVER DECIMAL(3,2), PACKTYPE CHAR(15), PACKCOST DECIMAL(5,2) ) Creates an empty table with 5 columns as defined.

Column name

Column data type

DROP TABLE Computer

Removes an existing table of given name. Name of the table to

be removed

Name of the table to be created

CREATE TABLE II (attribute based integrity constraint)

CREATE TABLE COMPUTER ( COMPID DECIMAL(2) NOT NULL, MFGNAME CHAR(15) NOT NULL, MFGMODEL CHAR(25), PROCTYPE DECIMAL(7,2) )

Integrity constraint (of a column atributu). In this case it says that

the value of the respective column is mandatory

When inserting a new row to the table, we need not specify values of all columns. Cells with missing values will get assigned a special value NULL. However, if integrity constrain NOT NULL, the value has to be specified mandatorily. If not, DBMS rejects execution of the command and an exception will be raised in the client program.

CREATE TABLE Films ( CODE CHAR(5) CONSTRAINT Firstkey PRIMARY KEY

TITLE VARCHAR(40) NOT NULL, DateProd DATE, KIND VARCHAR(10), LEN INTERVAL HOUR TO MINUTE )

Integrity constrain may be have got a name

CREATE TABLE II (attribute based integrity constraint)

CREATE TABLE IV (table based integrity constraint)

CREATE TABLE Films ( TITLE VARCHAR(40) NOT NULL, DateProd DATE, KIND VARCHAR(10), LEN INTERVAL HOUR TO MINUTE, CONSTRAINT pk_const PRIMARY KEY (TITLE, DateProd)

)

Integrity constrain on the table level

If the primary key consists of multiple attributes, we can not expres it by an attribute based integrity constraint as none of the attributes is primary key. The cpnstraint on primary key has to be expressed by a table-based integrity constraint in such a case. In the example above, the primary key is formed by a pair (TITLE, DateProd). PRIMARY KEY je one of possible table-based integrity constraints.

CREATE TABLE V (integrity constraints) CREATE TABLE osoby ( os_cislo NUMBER(5) NOT NULL, rod_cis VARCHAR2(30) NOT NULL UNIQUE, jmeno VARCHAR2(30) NOT NULL, prijmeni VARCHAR2(30) NOT NULL, plat NUMBER(5) CHECK ( plat > 5000 AND plat < 20000 ), cislo_prac NUMBER(5) NOT NULL, CONSTRAINT pk_osoby PRIMARY KEY ( os_cislo ), CONSTRAINT fk_prac FOREIGN KEY ( cislo_prac ) REFERENCES pracoviste ( cislo_prac ) ON UPDATE CASCADE ON DELETE CASCADE )

1.  Multiple integrity constrains may be specified for a column at the same time. 2.  Integrity constrain may have a form of a generic condition 3.  Any integrity constrain of a column may be expressed in terpms of an integrity constrain on a table level. 4. Referential integrity

1

2

3

4

CREATE TABLE VI (integrity constraints)

CREATE TABLE COURSES ( code VARCHAR2(10) PRIMARY KEY, name VARCHAR2(30) NOT NULL, credits NUMBER(2) DEFAULT 2, );

Default value of an attribute: If a row will be inserted that does not expressed a value of the „credits“ attribute, its value will not be NULL but 2, as it is its default value (see the CREATE TABLE statement above).

CREATE TABLE VII (generated ids)

CREATE SEQUENCE distrib_prim; CREATE TABLE distributors ( did integer PRIMARY KEY DEFAULT nextval('distrib_prim'), name varchar(40) NOT NULL CHECK (name <> '') )

1.  Definition of a sequence that will be named distrib_prim in given case. 2.  When inserting a new row (without specifying value for did), its default value will be

evalueted by nextval function with the nase of the sequence as an attribute.

1

1

This particular case is not an SQL standard. It is a syntax of DBMS PostgreSQL. Generated values standardized in SQL2006.

REFERENTIAL INTEGRITY CREATE TABLE employee ( emp_num NUMBER(5) PRIMARY KEY, ssn VARCHAR2(30) NOT NULL UNIQUE, dept_id NUMBER(5) NOT NULL, CONSTRAINT fk_dept FOREIGN KEY ( dept_id ) REFERENCES pracoviste ( dept_id ) ON UPDATE CASCADE ON DELETE CASCADE )

1. Integrity constraint fk_dept says, that attribute dept_id is a foreign key pointing to such a row of the department table, whose primary key’s value equals to the value of the dept_id attribute.

2. It means that rows representing all persons working in the given department have the same value of the dept_id column that refers to their common department.

3. What happens if somebody changes the value of primary key of the departments table? All the rows belonging to this department would suddebly point to a non-existing department. This is what should not happen. ON UPDATE ... section specify, how to solve this problem. In the particular case thsi section says ON UPDATE CASCADE. This means that the respective depertment primary key will be modified and in parallel, the value of foreign key in all rows of table employee that point to this department will be updated with the new value of the primary key.

REFERENTIAL INTEGRITY II

4. What happens if somebody deletes a row of the department table that represents a

common deprtoment of one or more employees? Rows representing these employees would suddenly point to a non-existing depertment. This is what should not happen. The ON DELETE … section specifies how this problem shall be handled. In our particular case this section specifies ON DELETE CASCADE. This means that the respectice department will be deleted. However, all its employees will be deleted in parallel.

CREATE TABLE employee ( emp_num NUMBER(5) PRIMARY KEY, ssn VARCHAR2(30) NOT NULL UNIQUE, dept_id NUMBER(5) NOT NULL, CONSTRAINT fk_dept FOREIGN KEY ( dept_id ) REFERENCES pracoviste ( dept_id ) ON UPDATE CASCADE ON DELETE CASCADE )

REFERENTIAL INTEGRITY III

What are options other than CASCADE?

1. RESTRICT – the modification that would violate the referential integrity will not be carried out. DBMS rejects execution of the modification of the primary key (ON UPDATE) or deletion of the row (ON DELETE) – an exception will be thrown.

2. SET NULL – the modification of the depertment table will be executed, but value of the foreign key of all rows of employee table that pointed to the deleted depertment will be set to NULL. It means they will not point to the respective department any more.

3. SET DEFAULT – similar as SET NULL – applicable if the foreign key has defined a

default value. Modifikátory CASCADE, RSTRICT, SET NULL, SET DEFAULT se v sekcích ON UPDATE a ON DELETE nastavují nezávisle.

CREATE TABLE employee ( emp_num NUMBER(5) PRIMARY KEY, ssn VARCHAR2(30) NOT NULL UNIQUE, dept_id NUMBER(5) NOT NULL, CONSTRAINT fk_dept FOREIGN KEY ( dept_id ) REFERENCES pracoviste ( dept_id ) ON UPDATE CASCADE ON DELETE CASCADE )

INSERT INTO

INSERT INTO EMPLOYEE VALUES ( 611, 'Dinh Melissa', 2963 ) List of columns not expressed – values will be assigend to columns in the order in which the columns have been defined in the CREATE TABLE statement. INSERT INTO EMPLOYEE ( EMPNUM, EMPNAME )

VALUES ( 611, 'Dinh Melissa' ) The section VALUES specifies one or more (comma separated) tuples of values. The values will be assigend to columns in the order given by the list of columns. Columns not introduced in the column list will get the NULL value. In this case EMPPHONE will get NULL.

Table name List of values

Table name

List of values

List of columns

SELECT I

SELECT <list of columns or *> FROM <relation definition> WHERE <selection condition> GROUP BY <list of columns>

HAVING <group filtering condition> ORDER BY <list of column_defs> column_def ::= <column name> [<asc|desc>]

Logical operators:

= equals <= less than or equal < les than >= greater than or equal > greater than <> not equal != not equal

SELECT II

PACKAGE table PACKID PACKNAME PACKVER PACKTYPE PACKCOST AC01 Boise Accounting 3.00 Accounting 725.83 DB32 Manta 1.50 Database 380.00 DB33 Manta 2.10 Database 430.18 SS11 Limitless View 5.30 Spreadsheet 217.95 WP08 Words & More 2.00 Word Processing 185.00 WP09 Freeware Processing 4.27 Word Processing 30.00 SELECT PACKID, PACKNAME, PACKCOST FROM PACKAGE WHERE PACKCOST >= 200 AND PACKCOST <= 400

SELECT PACKID, PACKNAME, PACKCOST FROM PACKAGE WHERE PACKCOST BETWEEN 200 AND 400

Result: PACKID PACKNAME PACKCOST DB32 Manta 380.00 SS11 Limitless View 217.95

SELECT III

PACKAGE table PACKID PACKNAME PACKVER PACKTYPE PACKCOST AC01 Boise Accounting 3.00 Accounting 725.83 DB32 Manta 1.50 Database 380.00 DB33 Manta 2.10 Database 430.18 SS11 Limitless View 5.30 Spreadsheet 217.95 WP08 Words & More 2.00 Word Processing 185.00 WP09 Freeware Processing 4.27 Word Processing 30.00

boolean predicate LIKE character % is a wildcard, i.e. matches with any charatcter (sub)string SELECT PACKID, PACKNAME FROM PACKAGE WHERE PACKNAME LIKE '%&%'

Result: PACKID PACKNAME WP08 Words & More

SELECT IV

Precate "IS NULL" is equal to „true“ iff the respective coulmn has assigned no value SELECT EMPNUM, EMPNAME

FROM EMPLOYEE WHERE EMPPHONE IS NULL

TRUE for those rows in which EMPPHONE has an undefined value, i.e. NULL.

SELECT V (arithemetic operators )

PACKAGE table PACKID PACKNAME PACKVER PACKTYPE PACKCOST AC01 Boise Accounting 3.00 Accounting 725.83 DB32 Manta 1.50 Database 380.00 DB33 Manta 2.10 Database 430.18 SS11 Limitless View 5.30 Spreadsheet 217.95 WP08 Words & More 2.00 Word Processing 185.00 WP09 Freeware Processing 4.27 Word Processing 30.00

SELECT PACKID, PACKNAME, ( .90 * PACKCOST ) FROM PACKAGE Result: PACKID PACKNAME EXP1

AC01 Boise Accounting 635.25 DB32 Manta 342.00 DB33 Manta 387.16 SS11 Limitless View 196.16 WP08 Words & More 166.50 WP09 Freeware Processing 27.00

Name of this column generated by DBMS client

i.e. 0.9 * 725.83 0.9 * 380.00 0.9 * 430.18 atd.

SELECT VI

PACKAGE table PACKID PACKNAME PACKVER PACKTYPE PACKCOST AC01 Boise Accounting 3.00 Accounting 725.83 DB32 Manta 1.50 Database 380.00 DB33 Manta 2.10 Database 430.18 SS11 Limitless View 5.30 Spreadsheet 217.95 WP08 Words & More 2.00 Word Processing 185.00 WP09 Freeware Processing 4.27 Word Processing 30.00 SELECT PACKID, PACKNAME, PACKTYPE FROM PACKAGE WHERE PACKTYPE IN ('Database', Spreadsheet', 'Word Processing')

SELECT PACKID, PACKNAME, PACKTYPE FROM PACKAGE WHERE PACKTYPE = 'Database' OR PACKTYPE = Spreadsheet' OR PACKTYPE = 'Word Processing'

Result: PACKID PACKNAME PACKTYPE DB32 Manta Database DB33 Manta Database SS11 Limitless View Spreadsheet WP08 Words & More Word Processing WP09 Freeware Processing Word Processing

SELECT VII (sorting)

PACKAGE table PACKID PACKNAME PACKVER PACKTYPE PACKCOST AC01 Boise Accounting 3.00 Accounting 725.83 DB32 Manta 1.50 Database 380.00 DB33 Manta 2.10 Database 430.18 SS11 Limitless View 5.30 Spreadsheet 217.95 WP08 Words & More 2.00 Word Processing 185.00 WP09 Freeware Processing 4.27 Word Processing 30.00 Order of rows in the result of the query is undefined unless specified by OREDR clause:

SELECT PACKID, PACKNAME, PACKTYPE, PACKCOST

FORM PACKAGE ORDER BY PACKTYPE, PACKCOST DESC

Rows will be oredered primarily by PACKTYPE. The order of rows with an equal value of PACKTYPE will be defined by PACKCOST. DESC ... descending ASC ... ascending (default) Result: PACKID PACKNAME PACKTYPE PACKCOST AC01 Boise Accounting 725.83 DB33 Manta Database 430.18 DB32 Manta Database 380.00 SS11 Limitless View Spreadsheet 217.95 WP08 Words & More Word Processing 185.00 WP09 Freeware Processing Word Processing 30.00